Ekonomia nr 42/2015 135

Determinants of Corruption in Ukrainian

Regions: Spatial Analysis*

Anastasiya Penska**

Abstract
This study investigates the determinants of corruption in Ukrainian
regions and applies spatial analysis to examine the relationship be-
tween corruption perception index and economic, political and cul-
tural factors. The results of the study show that economic growth
decreases corruption in the regions. The paper also challenges the
existing academic studies, as it shows positive impact of the number
of civil organizations in Ukrainian regions on corruption. The impli-
cations of the empirical research are, therefore, a valuable asset to the
existing literature on corruption. They indicate that dealing with cor-
ruption in Ukraine requires consideration of the influence of region-
specific corruption determinants.

Keywords: corruption, spatial analysis, economic growth, civil or-

ganizations
JEL Code: D73
DOI: http://dx.doi.org/10.17451/eko/42/2015/123

*
 his article is based on the master thesis written under supervision of Katarzyna Metelska-Szaniaw-
T
ska, PhD, at the Faculty of Economic Sciences, University of Warsaw.
**
Faculty of Economic Sciences, University of Warsaw.
136 Anastasiya Penska

Acknowledgements

This paper would not be possible without the kind guidance of Katarzyna Metelska-
-Szaniawska, PhD. I have learned so much about being a researcher from her, she
would always suggest fresh perspective to solving research problems. I am there-
fore genuinely grateful for her contribution to the given paper. The contribution
of Kyiv Institute of Sociology also cannot be left unnoted as I was provided with
extensive data necessary for the study.
I would also like to express my gratitude to WNE UW professors for provid-
ing me with the knowledge and skills without which I would not be able to conduct
the research. I would finally like to thank my husband for emotional support and
kind wise words that kept me motivated.
 Ekonomia nr 42/2015 137

1. Introduction

Transparency International (2009) defines corruption as abuse of entrusted power

for the private gain. While this definition is vague and general, corruption has been
an issue of public debate and academic research for many years. Some authors
openly blame corruption for undermining operation of markets and therefore re-
sulting in slower economic development (Barro 1997; Bai et al. 2013). Others are
less convinced whether corruption is entirely a negative phenomenon and point
out that corruption may help to avoid bureaucracy and even stimulate some eco-
nomic growth (Leff 1964).
Since its independence Ukraine has been challenged by widespread corruption
activities which put in danger its economic and political future. According to re-
port published by the European Research Association, USAID, and Kyiv Interna-
tional Institute of Sociologyin 2011, 10% of Ukrainian population is reported to be
forced into corruptive activities, with courts and universities ranked with the high-
est perceived corruption – widespread values of 60.5% and 47.1%, respectively
(European Research Association [ERA] et al. 2011).
Country’s Soviet past and geographic location often become the green light for
corruption in Ukraine. Moreover, current economic and political situation in the
country fuelled by the military conflict in the East of the country leave little room
for anticorruption measures. In fact, local population seems to be used to corrup-
tion being part of their lives while people also do not believe in future without cor-
ruption. In 2011 low trust levels in both local and state government were registered
with only 20.1% of population trusting local authorities, 14% having trust in the
President and only 7% in judicial system (ERA et al. 2011).
The motivation for this study lies in the importance of corruption issue for
Ukraine’s further development, economically, politically and culturally. The aim
of the paper is to determine which factors influence corruption perception index
in Ukrainian regions and developing policy suggestions that may be used for the
bottom-up anti-corruption reforms. The goal of this paper is, therefore, to apply
spatial analysis techniques in order to investigate economic, political and cultur-
al determinants of corruption perception in Ukrainian regions. Hence, the added
value of the given paper is in empirical analysis of CPI (Corruption Perception In-
dex) determinants in spatial regional dimension. Today there is a lack of a similar
studies on the Ukrainian regions; this work drives methodological and empirical
experience from other studies to come up with suggestions and recommendations
that may be utilized to decrease corruption in Ukraine.
Mainly, this research strives to verify three hypotheses: First one states that in-
crease in GRP (Gross Regional Product) growth per capita and increase in real in-
come per capita have a negative effect on CPI. Second hypothesis predicts that in-
crease in number of civil organizations in a region and higher regional integration
138 Anastasiya Penska

index decrease values of CPI, while third hypothesis suggests that higher number
of MPs (Member of Parliament) originating from a given region has a positive
impact on CPI. The paper uses regional data and spatial analysis and Two-Stage
Least Squares technique with instrumental variables to capture the impact of vari-
ous determinants on CPI.
The paper is divided into three parts. The first one features literature review,
then the second one proceeds with data description and methodology, while the
third presents the estimation of empirical model and its results. The final section
of the paper provides concluding remarks and suggests further steps to advance in
fighting corruption in Ukraine.

2. Literature review
This section extensively covers existing literature on the matter of corruption anal-
ysis. Particularly, my interest here is in covering the literature on defining corrup-
tion and determinants of corruption, as those areas of study which are critical for
my research. I firstly define certain groups of determinants as economic, political
and cultural and then focus on the very determinants supported by the relevant
literature. Although academic literature includes a considerable amount of studies
concerning the determinants of corruption, not all of them are applicable to the
regional dimension research that is conducted in this paper. For this reason, only
determinants relevant to regional dimension will be reviewed.

2.1. Defining corruption

As mentioned earlier, it is difficult to define and measure corruption as it is both
illegal and hidden. Majority of empirical studies use CPI that is annually published
by Transparency International (Bernalez 2008; Alam 1995; Husted 1999; and oth-
ers). The index is based on ten international surveys of the perceptions of business
people and country experts regarding corruption around the world and is published
annually. However, the popular critique of such approach claims that due to its in-
trinsic nature, corruption is unique in its form and appearance in each country and
therefore adopting a general index evaluating many countries at the same time may
result in biased results (Fiorino and Galli 2010, 2012). As the argument goes, each
country has specific and peculiar features meaning that investigation of corruption
should take into account cultural and legal differences of the country.
The abovementioned debate results in widespread doubts regarding the meas-
ure that should be chosen when examining corruption. For instance, Fiorino and
Galli (2010) use number of crimes committed by the public officials as dependent
 Ekonomia nr 42/2015 139

variable. This approach, however, has some significant drawbacks that are also
admitted by the authors. Mainly, the problem lies in the fact that registered perse-
cutions for the crimes committed by the public officials only captures ‘revealed’
corruption, while omitting the ‘hidden’ one. Moran further specifies that corrup-
tion captures state-society relations, political systems, development trajectories
and types of external linkages (Moran 1999). Indeed, the determinants influencing
corruption are largely agreed to be broadly divided into economic, political and
cultural.

2.2. Determinants of corruption

It is important to define determinants of corruption in order to proceed with the
investigation of CPI. Hence, a closer look at the economic, political and cultural
determinants of corruption is taken as well as on the literature that uses those de-
terminants to study corruption. First, the relevance of the group of determinants is
described and the determinants themselves are discussed.

Economic determinants
Corruption determinants referred to as economic are grouped by assumption that
state intervention is conducted not through the bureaucratic system and legal ways
of impact but using administrative mechanisms for private purposes causing cor-
ruption (Del Monte and Papagni 2007). Shabbir and Anwar (2007) also find that
economic factors have more effect on corruption than non-economic ones when
investigating this issue for 41 developing countries. This argument is generally
supported by the liberal economists who believe that corruption may be fought
only when minimising the role of the government in the economic life of the coun-
try. Otherwise, corruption is largely used as an instrument of control and influence
on the government by the interest groups.
The role of interest groups is associated with two-way causality between cor-
ruption and economic activity. On the one hand, interest groups often possess cer-
tain economic power, which allows them to influence the ruling party, as they get
to avoid taxation, win tenders, or buy land cheaper than market prices. On the oth-
er hand, autocratic elite controls interest groups by giving them access to corrup-
tion rent. This issue is commented by Hollyer (2011), who proves empirically that
elites may systematically manipulate access to rent to provide incentive for higher
performance for both local and state officials. The author explains that autocratic
elite controls the interest groups by granting career opportunities to best perform-
ers and punishing the inefficient officials. At the same time, according to Hollyer
140 Anastasiya Penska

(2011) if elite is supported by interest groups acting in the name of their ideologi-
cal empathy, there may be lower corruption in the country. Economic development
is therefore claimed as both the tool for interest groups and aim for autocratic elite
which interact under corruption.
Some studies state that economic development serves as a determining factor
for the level of corruption in the country. Huntington (1968) and Myrdal (1968)
advocate the idea that at the early stages of development, societies face higher
levels of corruption, which decreases after certain development level is reached
and then decreases along with the pace of economic development. This phenom-
enon, as the argument goes, causes the inverted U-shaped relationship of economic
development and corruption (Treisman 2000). The finding of the U-shaped rela-
tion between corruption and economic growth gives useful interpretation of this
relationship for developing countries. It also allows us to speculate on the level of
economic potential of a given country by looking at the correlation sign of the cor-
ruption and economic growth. Hollyer (2011) challenges this idea and emphasizes
that in less developed countries where elites built their power on ideological affin-
ity, corruption could be relatively low and well controlled by the elite, as it was in
the Soviet Union.
Bai et al. (2013) examine corruption dependence on economic growth in Vi-
etnam. They build a model to examine whether in poor countries firms have an
opportunity to avoid corruption and bribes while stimulating economic growth.
Empirical study discovers that firms operating in multiple provinces and/or trans-
ferable land property rights experience less corruption as they are more flexible
(Bai et al. 2013). As the study shows, operating in various provinces not only
differentiates the risks of business expropriation as a result of avoiding bribes but
also redistributes the wealth between regions leading to lower income inequality.
Income inequality is another important economic determinant of corruption
and is used in many studies as corruption explanatory variable. Shabbir and Anwar
(2007) argue that income inequality makes population preoccupied with income
redistribution rather that honesty monitoring. As a result more people are prompt
to pay bribes, making corruption more socially acceptable. The authors also find
income inequality to be positively correlated with the level of corruption (Shabbir
and Anwar 2007). Gupta and Davoodi (1998) conclude that corruption is positively
correlated with income inequality, whereas policy aimed at decreasing corruption
also decreases income inequality and poverty. Alam (1995) and Johnston (1989)
study the mutual causality relation between income inequality and corruption that
corruption causes greater income inequality, while the inequality in income distri-
bution promotes higher levels of corruption.
Dzhumashev (2013) has conducted a study regarding two-way relationship
between governmental expenditures and corruption and their impact on economic
growth. His conclusions are that governmental expenditures trigger corruption,
 Ekonomia nr 42/2015 141

which in turn, leads to distortions in the structure of governmental expenditures.

Fiorino and Galli (2010) agree with the statement and claim that lower government
expenditures with fewer regulations are generally associated with lower corrup-
tion.
Sandholtz and Koetzle (2000) look at the way unemployment impacts the level
of corruption. They conclude that high marginal value of money in poor countries
makes bribing beneficial for both givers and takers of bribes. Unemployment is one
of the reasons why people are ready to pay bribes in order to open higher income
gain opportunities and employment. Buying and selling positions, as the case of
corruption, is claimed to be positively correlated with corruption. Alongside with
unemployment such economic determinant as inflation is also thought to increase
corruption, as it serves as another destabilising factor. Among others, Bai et al.
(2013) and Jung (1985) take inflation into account when investigating corruption.

Political determinants
Political factors influencing corruption are grounded in the costs which corrupted
officials face in pursue of personal benefit. Putnam (1993) defines political at-
tention as being a luxury good. Such characterization allows him to determine
factors influencing the ‘price’ that a politician faces when acquiring political at-
tention by means of corruption. Elections are perfect example of the corruption
encouraging environment (North 1971). Moreover, local elections are even more
corruption-prone as corrupt official can focus on specific sector rather than the
whole government, making corruption less risky. Whether an official has to pay
a high price for political attention depends on civic participation of his electorate.
Balancing between political representation of different interest groups in turn pro-
vokes more corrupt activities.
The level of civic and political participation of the population highly depends
on the level of education. This concept is central in research by Lipset (1960) and
Glaeser et al. (2004) who examine the way education, political and civic engage-
ment influence corruption. The authors come to conclusion that more highly edu-
cated people, as well as those active both politically and civically, have a tenden-
cy to monitor honesty, hence decreasing the level of corruption. Higher political
awareness also leads to stronger civic community (Del Monte and Papagni, 2007).
This idea is challenged by results of empirical study by Fiorino and Galli (2010)
who examine corruption in Italian regions. One of the interesting findings of the au-
thors is a positive correlation between corruption and education, which they explain
as the ability to bypass and evade regulation and that is only increased with higher
levels of education. Their study focuses on the impact of corruption on economic
growth. It has been found that economic growth is highly correlated with public
expenditures, effect of which is undermined by corruption (Fiorino and Galli 2012).
142 Anastasiya Penska

Higher level of democracy and political plurality is also claimed to decrease

corruption. As Del Monte and Papagni (2007) explain, higher political competition
influences corruption in two ways: First of all, it stimulates creation of new politi-
cal parties and therefore raises political awareness and secondly, it fragmentizes
electorate’s choice, therefore giving all political forces more equal chances and
decreasing need of corruptive activities at the same time.

Cultural determinants
This group of corruption determinants suggests that corruption largely depends on
social capital, which is associated with strong civil society (Mauro 1995; Fearon
and Laitin 1996; Wines and Napier 1992). They claim that corruption patterns tend
to correspond to the rules of the ‘legal culture’ (Fiorino and Galli 2010). Wines
and Napier (1992) conclude that dominant cultural norms influence social accept-
ance of corruption. Corruption tends to increase when the state laws are in disso-
nance with strongly integrated culture and well defined moral and cultural norms.
On the other hand, as Bernaldez (2008) and Shkurpat (2006) argue such disso-
nance increases the risk of corrupted activities for officials when public opinion
has a strong vision of honesty resulting in protests and revolutions – hypothesis
which has been empirically proved by recent events in Ukraine (especially after
the Revolution of Dignity of 2013, see Koshkina 2015).
According to cultural approach to corruption determinants, trust in govern-
ment is seen as a public good which impacts the perceived quality of life of the
population. Apart from integration level mentioned above, civil organizations are
believed to influence the revival of public ‘legal culture’ (Coleman 1990; Hun-
tington 1968). Academic research also suggests that higher integration level of the
population – meaning lower religious, language and ethnic fragmentation – leads
to more active participation and stronger moral norms, therefore decreasing cor-
ruption (Coleman 1990).
Senior (2006) puts together a thorough study of corruption in many countries
assessing their corruption level and comes to interesting conclusions regarding the
role of integration of population. The author uses CPI provided by Transparency
International as dependent variable, extensively analysing the determinants of cor-
ruption in his book The Big C: Cases, Causes, Consequences, Cures. Senior indi-
cates inverse relationship between corruption and social integration and observes
direct relationship of the dependent variable and government’s economic interven-
tions. He uses causality tests to discover complex interdependencies between cor-
ruption and economic variables and well as political ones.
Kappor and Ravi (2009) distinguish between corruption norms and govern-
ment effectiveness instead of investigating the use of corruption index. They find
 Ekonomia nr 42/2015 143

that corrupt behaviour of UN officials cannot be explained by corruption norms

alone and is largely dependent on government effectiveness. Kapoor and Ravi
(2009) suggest that cultural norms are not best explanation for corrupt behaviour
unless supported by data on government effectiveness, which they define as com-
posite indicator showing level of office organization, official’s education level,
quality of civil service and other factors influencing a given official.
The abovementioned determinants are not self-exclusive, moreover, it is neces-
sary to apply each of them to the country-case study of corruption. Such approach
allows us to explain as many interrelations between the variables as possible and
tailor the model to a specific country based on peculiar features and data availabili-
ty. This argument is used by Bernaldez (2014) who gathers into one study econom-
ic, political, cultural and organizational determinants. As he distinguishes between
cultural and organizational determinants his classification is different from the one
mentioned previously with some of them only relevant for international and not
country-level regional analysis. Bernaldez's (2014) work thus serves as a summary
platform of the interrelationships between corruption and its determinants which
may come useful for interpretation of empirical results.
It is important to note that while the abovementioned studies are relevant for
the given work, they do not take into account the distribution of spatial dimension
of corruption across the regions. This might be due to insignificance of spatial
autocorrelation in the model, or it may also relate to the fact that majority of stud-
ies examine rather the international aspect of corruption and do not focus on an
in-country regional level determinants.

2.3. Studies on Ukraine

There is little literature that would examine corruption in Ukrainian regions,
while there are no studies that take into account spatial dimension of corruption in
Ukraine.
Neutze and Karatnycky (2007) put together an empirical study that provides
a thorough insight into the corruption in Ukraine. It gives a generalised report on
the corruption in economic sector, judicial sector, executive and legislative sectors,
political corruption, as well as the role of private sector and media. The main con-
clusion of Neutze and Karatnycky (2007) is that Ukraine needs to ensure thorough
implementation of current anti-corruption laws, unite all political forces in passing
more corruption-fighting legislation, establish constructive dialogue between civil
society and the government, and also initiate anti-corruption programme in judiciary
and health sectors as the first priority. Neutze and Karatnycky (2007) also name gov-
ernmental expenditures watchdogs, civic education, increase MPs’ accountability
and empower civil organizations with ability to exercise pressure on the government.
144 Anastasiya Penska

The paper by ERA et al. (2011) features an extensive analysis of corruption

perception in Ukraine. The study based on the regional surveys indicates the lead-
ing causes of corruption that are perceived as such by the population. According
to the survey, officials’ desire to use their position in public office for personal
gain is the main reason for corruption, which echoes political theory of corruption
determinants. The subsequent factors – inadequate control of the law enforcement
over officials and absence of strong political will of the highest levels of the gov-
ernment – also point at political reasons for corruption in Ukraine.
However, if one wants to investigate regional level of corruption perception,
it is important to look at the areas in which people interact with corruption the
most. According to the ERA, USAID and Kyiv International Institute of Sociol-
ogy (2011) paper, top areas where population faces corruption are: healthcare,
universities, business regulation and inspection, state auto inspection, state subsi-
dised housing and governmental jobs. The study, however, concludes that extor-
tion to bribery in not equal across the regions – it is higher in eastern and southern
regions and lower in western and central regions.
Vyshnyak (2009) explains the differences between corruption spread across
Ukraine by looking at regional cultural, ethnical and language differences across
the regions. She conducts an empirical study examining the amount of people
speaking Ukrainian, Russian and other languages in different life spheres like job,
home or university. Generally, the author concludes that Ukrainian speaking re-
gions report higher level of social integration and political participation (Vyshnyak
2009).
USAID (2006) report suggests that the cleft between East and West of the
country fuels political crisis in Ukraine, as politicians use and reinforce the dif-
ference between the regions to acquire larger electorate. This argument brings us
back to the cultural determinants of corruption, once more pointing out at the need
of taking into account the spatial dimension of corruption in Ukraine.
Having analysed the literature concerning the definition of corruption as well
as the determinants of corruption, the following part proceeds with data description
and methodology used for the empirical research. The consequent section features
estimation of the model and its results. Finally, the paper also presents conclusions
on the conducted research.

3. Data description and methodology

The study presented in the paper investigates the determinants of corruption in
Ukrainian regions and looks at economic, political and cultural factors influenc-
ing CPI at the regional level. Ukraine is divided into 25 administrative units: 24
regions (‘oblasts’ in Ukrainian) and Autonomous Republic of Crimea. The given
 Ekonomia nr 42/2015 145

research focuses on the time period of 2004−2011 and uses annual data for 25
regions of Ukraine. After Orange Revolution in 2004 Ukraine started taking its
first steps towards stronger civil society and lower corruption (Koshkina 2015). As
a part of this process Ukrainian Statistical Bureau (USB) reformed its statistical
methods and made the information available to the wide public on its website (Ko-
shkina 2015). Furthermore, statistical data became even more influenced by exter-
nal forces with beginning of the military conflict with Russia. Studying corruption
determinants during 2004−2011 is therefore the most suitable time period for both
unbiased results with the largest possible data sample.
The following equation is estimated:
(1)

for i = 1,..., 25 and t = 2004,..., 2011, where X is a vector of explanatory vari-
ables and ϵ is the error term.
More specifically, the following spatial model was constructed:
(2)

(3)

Where X is the vector of explanatory variables; ρ is the coefficient estimated

for the spatial lag; W is the spatial weights matrix; λ is the coefficient estimated for
the spatial error; u is the unobserved non-spatial error for every observation.

3.1. Dependent variable

Corruption perception index (variable: corruptind) is used as dependent variable
for this study. The concept of this determinant has been developed by Transpar-
ency International in 1995 and uses international surveys to create a score for each
country on the scale of 0 to 100 showing how corrupt its public sectors are per-
ceived to be (Transparency International, 2014). Transparency International uses
empirical data submitted by the supporting national and international institutions
to produce the index with 0 standing for the country perceived as highly corrupt
to 100 which means no corruption.
Following the guidelines of Transparency International, Ukrainian statisti-
cal institutions took the approach of corruption perception index to regional level
to assess the perception of corruption and its dynamics in every oblast in Ukraine.
Kyiv International Institute of Sociology and in the framework of USAID funded
project called “The Ukraine National Initiatives to Enhance Reforms” gathered the
index data for the years 2007‒2011 (ERA et al. 2011) and produced regional CPI
for Ukrainian oblasts and Autonomous Republic of Crimea. The index varies from
146 Anastasiya Penska

0 to 100, with 0 standing for no corruption and 100 for very corrupted region as
perceived by the population. This CPI measure is used in the given paper to inves-
tigate corruption determinants in Ukrainian regions.
The choice of CPI as dependent variable to measure level of corruption in a re-
gion has a number of reasons. First of all, although the index is a subjective indica-
tor, it reflects everyday realities of the Ukrainian population and therefore can be
considered a good proxy for the real level of corruption in the country. Secondly,
the index has been created taking into account both Ukrainian realities and Trans-
parency International methodology. On the one hand, corruption perception might
to some extent be higher than actual corruption interaction (ERA et al. 2011), but
on the other hand, the index indicates to what extent corruption is accepted as a part
of social reality in Ukraine, potentially complicating the process of fighting it.

3.2. Explanatory variables

The study presented in this paper uses the following explanatory variables:
Gross regional product growth per capita (variable: grpgr) has been includ-
ed as an explanatory variable for the study. USB Yearbooks provide measurements
of GRP with the Gross Value Added (GVA) that is further defined as the difference
between output and intermediate consumption, which includes the primary income
generated and distributed by producers.
Using real income per capita (variable: realincomepc) as an explanatory vari-
able is a good method of capturing the way real income of the population influenc-
es perception of corruption. This variable shows real average salary in the region.
The variable primarily affects the perception part of the CPI as lower salaries are
believed to make people blame corruption.
Another important variable that will be used to explain corruption is unemploy-
ment rate (variable: unem). The level of unemployment is an important determinant
of CPI as it reflects the willingness and need to pay bribes. The analysis by Sandholtz
and Koetzle (2000) suggests that higher unemployment rate encourages population
to look for illegal employment and pay bribes to obtain employment. Variable is pre-
sented in percentage of unemployment growth rate compared to the base year 2003.
The money allocated to each region is captured by using governmental ex-
penditures (variable: govspend) as an explanatory variable. This variable shows
the amount of money allocated to a given region from central government budget.
Crime rate (variable: crime) is also used to investigate CPI. The variable
shows overall number of registered crimes in the region and is assumed to cor-
relate with the amount of crimes committed by governmental officials, as those are
not explicitly recorded. As Dzhumashev (2013) shows it, there is positive relation-
ship between governmental spending and CPI.
 Ekonomia nr 42/2015 147

Another variable suggested by the analysed academic literature is the number

of MPs in the current parliament (variable: mp). Fiorino and Galli (2010, 2012)
use number of legislators from a given region to control for region’s interests rep-
resentation in the parliament. It is assumed that in a relatively corrupted country,
legislators have strong connections in their home regions, and once they get to the
parliament those connections gain momentum using the legislator’s power.
Same logic is used to apply number of political parties (variable: parties) as
an explanatory variable for the empirical study. This variable is a proxy for political
plurality in the region. Del Monte and Papagni (2007) argue that political plurality
is an important determinant of corruption as the higher is competition for electorate
between political parties, the lower their incentive to use corruption to acquire vot-
ers. The authors claim this relationship is directly related to the number of political
parties available in the analysed territory (Del Monte and Papagni 2007).
Number of civil organizations (variable: civilorg) is very useful when captur-
ing the impact of civil society on corruption perception. The variable shows the
number of registered civil organizations in the region and is used as a proxy for
civil participation in the region.
Integration index (variable: integration) is also used to evaluate the influence
of civil society on CPI. Since the regional data is spatially autocorrelated, it is
logical to take into account the peculiarities of each region. It is shown by Hus-
ted (1999) that these peculiarities are correlated with corruption index. Integration
index is a complex indicator produced by USB and composed of such indicators
as social wellbeing, appropriate workplace, linguistic and ethnic coherence and
population growth. The data used for the index includes both state-level statistics
and data from surveys. Integration index supports the hypothesis of cultural theory
of corruption by providing the insight of the region’s unity and coherence.

3.3. Preliminary analysis

Descriptive statistics of variables used in the model have been presented below and
the correlations between the variables have been looked into.
The sign of the correlation between CPI and GRP growth per capita and real in-
come is negative, as expected and mentioned in the literature. There is also relative-
ly strong positive correlation between real income and civil organizations as well as
between number of civil organizations and integration. The most unexpected sign
of the correlation matrix is positive correlation between CPI and civil organizations
variable. The analysed literature predicts that civil organizations should decrease
corruption via increasing the political and civil participation as well as educate peo-
ple about the ways of fighting corruption. This relationship is particularly inter-
esting to investigate as the correlation matrix also shows that the number of civil
148 Anastasiya Penska

organizations rises as real income increases. This may indicate that civil organiza-
tions are inefficient in fighting against corruption, but it is important to estimate the
model to prove this suggestion. Such statement is also supported by strong positive
correlation between number of civil organizations and number of crimes.
The analysed literature indicate an endogeneity problem in the sample. The
most commonly used method of dealing with this problem is using instrumental
variables. This study uses three instrumental variables: number of student gradu-
ates, fixed capital investment and level of urbanization. The choice of the variables
is based on the analysed academic literature (Pigliaru 2009; Olsen 1996; Tanzi
and Davoodi 1996; Lipset 1960; Glaeser et al. 2004; Fiorino and Galli 2012). In
particular, the following instrumental variables will be used:
Number of student graduates from higher education institutions (variable:
stud) is used as one of the instrumental variables. Fiorino and Galli (2010, 2012) and
Lipset (1960) and Glaeseret al. (2004) argue that well educated people show higher
rate of civil and political participation. On the other hand, in corrupted environment
clever minds might fall into corrupt activities. Somewhat controversial, but number
of graduates might be a convenient instrument for such indicators as civic organiza-
tions and level of crime in the regions when used as a proxy for human capital.
Fixed capital investment (variable: inv) is another instrumental variable that
is used in the model. This variable is used as a proxy for capital. It is defined
by Statistical Yearbook of Ukraine as including: “expenditures for capital con-
struction (new construction, including the expansion of the operating enterprises,
buildings and installations; their technical re-equipment and reconstruction; main-
tenance of production capacities); expenditures for the purchase of machinery and
equipment without capital construction”. The influence of the fixed capital invest-
ment on corruption is complex, as on the one hand, corrupted activities may be
attracted by investments in the region, while on the other hand, investment rich
region might attract new corruption.
Urbanization level (variable: urbanization) has been used as an instrument
in the model. Urbanization level shows proportion between urban and rural popu-
lation in the region. The measure of urbanization level is a ratio between urban
population and rural population in a region for the given period of time. Level of
urbanization of a given region is determinant of the level of corruption as corrup-
tion is more widespread in cities, as it is argued by Glaeser et al. (2004).

4. Estimation and results

This model for spatial autocorrelation was tested both globally and locally and
then a choice has been made between spatial lag and spatial error model specifi-
cation. Having finalized the model, the author has estimated and performed post-
 Ekonomia nr 42/2015 149

estimation, under-identification, homoscedasticity, normal distribution of residuals

test and Sargan test of overidentifying restrictions.

4.1. Final model specification

The data are strongly balanced without gaps, that is, there are 200 observations
for 25 regions for the years 2004−2011. Next step is to specify a weighting ma-
trix in a way to capture spatial effect most efficiently. The simplest method is just
to capture the effect of common border. However, this specification does not take
into account the information on how far one region is from another. Therefore, the
farther two regions are from each other, the smaller the weight that shows the com-
mon variation in their variables. As in Ukrainian administrative centres which also
play the role of economic centres, distance between them in kilometres was used
as a proxy for distance. The best weights are the inverses of the distance squared
(Kamarianakis and Le Gallo 2003).
In order to apply spatial analysis the first step is to test whether the variables in
the model are globally spatially autocorrelated. According to Moran I test, variable
is positively spatially correlated when I>E (I) and negatively spatially autocorre-
lated when I<E (I), where I stands for Moran’s global index of spatial autocorrela-
tion and E (I) stands for the expected value of I (Pisati 2012). Table 1 shows that
indeed all of the analysed variables show positive global spatial autocorrelation,
meaning that there is overall degree of similarity between spatially close regions.
The p-values for all variables are lower than 5% significance level; therefore the
null hypothesis of no global spatial correlation can be rejected and it can be con-
cluded that all variables show significant global spatial autocorrelation.

Table 1. Measures of global spatial autocorrelation

Variable I E (I) p-value
Corruptind 0.001 -0.006 0.005*
Grpgr 0.000 -0.006 0.027**
Realincomepc 0.016 -0.006 0.002*
Unem 0.075 -0.006 0.000*
Govspend -0.003 -0.016 0.017**
Crime 0.253 -0.006 0.000*
Mp -0.040 -0.006 0.000*
Parties 0.009 -0.006 0.023**
Civilorg 0.059 -0.006 0.000*
Integration -0.033 -0.006 0.000*
*
significant at p=0.01, ** significant at p=0.05
Source: own calculations.
150 Anastasiya Penska

After it has been proven that all variables in the model show positive global
spatial autocorrelation, it is important to perform local spatial autocorrelation test
to see which regions have stronger spatial patterns (Table 2).The result of the test
will show whether region ri is surrounded by regions that, on average, are similar
to region ri.

Table 2. Local spatial autocorrelation test for corruption perception index

Real
Corrup- Govs- Par- Civi- Integra-
Region Grpgr inco- Unem Crime Mp
ind pend ties lorg tion
mepc
A.R. Crimea
Vinnytsya + + +
Volyn + + + +
Dnipro-
+ + + +
petrivsk
Donetsk + + + + + + + +
Zhytomyr + + + + + +
Zakarpatya + + + + +
Zaporizhya + + + + +
Iv.Frankivsk + + + + + +
Kyiv + + + + + + +
Kirovograd +
Luhansk + + + + +
Lviv + + + + +
Mykolaiv
Odessa + + + + +
Poltava + + + +
Rivne + + + + + +
Sumy + + + +
Ternopil + + + + + + + +
Kharkiv + + + + + + +
Kherson
Khmelnyt-
+ + +
skyy
Cherkasy
Chernivtsi + + + + + + + +
Chernihiv +
Source: own calculations.

Local spatial autocorrelation test has been performed (Table 2). Similarly,
global spatial autocorrelation test, Moran local spatial autocorrelation index I and
 Ekonomia nr 42/2015 151

its expected value E (I) are calculated to determine whether a region is a ‘hot spot’
(I>E (I)) or a ‘cold spot’ (I<E (I)) in terms of similarity of the neighbouring re-
gions (Pisati 2012). As the test calculates index for each region for each year of the
sample, it is more reasonable to indicate which regions perform as ‘hot spots’ for
each variable. Statistically significant (p<0.10) ‘hot spot’ regions are marked with
a “+” in Table 2. Regions which are not marked with “+” do not perform as ‘hot
spots’ for a given variable. The regions are listed in alphabetical order.
There are several regions that perform as hotspots for almost all of the ana-
lysed variables (Table 2). It is possible to draw several clusters of regions with the
‘hot spots’: Western cluster with Lviv, Zakarpatya, Ivano-Frankivsk, Ternopil and
Chernivtsi as the most influential regions, Eastern cluster with Donetsk, Kharkiv
and Luhansk regions as the ‘hot spot’, and Southern cluster featuring Odessa as the
leading region. Central region is led by Kyiv and Northern part of Ukraine seems
to be rather fractured with no distinctly seen cluster.
The next step in estimating the model is choosing between spatial lag and spa-
tial error model. The spatial diagnosis test determines whether spatial patterns area
result of spatial lag (when the outcome in one region is affected by the outcome in
the neighbouring ones) or they result from spatial error (the outcome in one region
is affected by unknown characteristics of the neighbouring regions). Results of the
test indicate that it is better to use spatial lag model, which means that change in
one variable in a given region is influenced by the change in that variable in the
neighbouring regions.

4.2. Model estimation and testing

Two-Stage Least Squares estimation with Instrumental Variables (2SLS (IV)) was
used. This estimation method is widely used in spatial models (Linderhof et al.
2011). As can be seen the impact of grpgr, realincomepc, civilorg and govspend is
positive and statistically significant at p<0.01 (Table 3). The influence of parties is
positive and integration affects CPI negatively at p<0.05. The variables crime, mp
and unem did not show significant impact on CPI.

Table 3. Estimation results of spatial model with 2SLS (IV) estimator

Variable 2SLS (IV)
Grpgr -0.916
0.001*
Realincomepc 0.551
0.001*
Unem -0.014
0.655
152 Anastasiya Penska

Variable 2SLS (IV)

Govspend 0.164
0.000*
Crime -0.192
0.178
Mp 0.015
0.725
Parties 0.198
0.023**
Civilorg 0.452
0.013*
Integration -0.439
0.048**
R2 0.608
*
p<0.01, ** p<0.05
Source: own calculations.

The post-estimation test results imply rejection of under-identification of equa-

tion at 1% which means that the instrumental variables were chosen correctly. The
author has also performed Sargan test, homoscedasticity test and Ramsey/Pesaran-
Taylor RESET test. Weak identification test was passed and results of Sargan test
of over-identifying restrictions do not give reason to reject the null hypothesis that
instruments are valid. There is no reason to reject homoscedasticity hypothesis and
residuals are normally distributed. Ramsey/Pesaran-Taylor RESET is also passed,
so there is no reason to reject the null that there are no neglected nonlinearities.
Estimation results imply that GRP growth per capita has a strong negative ef-
fect on CPI and is statistically significant. Real income per capita also performs
the way it was expected and has a significant strong negative effect on CPI. Thus,
outcome completely supports Hypothesis no. 1. Such outcome is overall predicted
by academic literature analysing economic determinants of corruption.
According to Table 3, neither the level of crime nor the number of MPs orig-
inating from a given region appeared to have statistically significant impact on
CPI. Therefore, Hypothesis no. 2 is not supported by empirical study. It is possible
to speculate about two reasons for this outcome. First, it is likely that number of
MPs is relative even across the country, which is also supported by Table 2 indicat-
ing that there is no clear ‘hot spot’ region for this variable. Secondly, the level of
crime might have appeared insignificant, as official statistics only represent a part
of crime that is involved in corruption, therefore causing biased results.
Hypothesis no.3 is not confirmed by estimation results as well. In fact, this
result proves that there exists a certain dissonance in the way civil organizations
influence corruption situation in Ukraine. The U-shaped relationship, predicted
by the literature (Treisman 2000), seems to indicate the country has passed the
 Ekonomia nr 42/2015 153

time when economic growth was encouraging corruption, but civil organizations
that are supposed to decrease corruption even further appear to be positively cor-
related with it. One reason why civil organizations may be associated with cor-
ruption is because they were established before the peak point of the U-shaped
relationship between corruption and economic growth was reached. In a sense,
increased number of civil organizations is triggered by the economic growth (as
it is also proved by the correlation matrix). However, it also might have happened
that in a corrupted country like Ukraine, which is actively encouraged by the Eu-
ropean community to build on its civil society, was not yet ready for the increasing
amount of civil organizations. As a result civil organizations became new sources
and breeding grounds for corrupt practices. While, on the one hand, they might
indeed have some positive influence on civil society, on the other hand, they may
also serve as a tool for money laundering.
It was also discovered that governmental spending has significant positive in-
fluence on CPI. This result is rather controversial, since I am looking at perception
of corruption rather than registered corruption activities. However, the estima-
tion results clearly indicate that regions which are allocated higher governmental
spending are associated with higher values of CPI. The explanation for this out-
come may lie in the fact that money allocated to a given region does not reach its
aim and ends up supporting corrupted activities and used for personal benefit of the
responsible for the distribution local officials.
Number of MPs born in a specific region and present in the current parliament
was found to be insignificant, as well as the crime rate. Insignificance of the first
variable is rather hard to explain as there is clearly unequal representation of the
regions in the parliament. However, it might be the case that although an MP was
born in a given region it does not necessarily mean his presence in the parliament
would benefit corrupted structures in that region, as it was proved by the estima-
tion. As for the second variable, crime rate appeared to be insignificant most likely
because the crimes associated with corruption are not registered in official statis-
tics. On the contrary, the relation between crime rate and corruption might be op-
posite, following this logic: As more bribes are paid, fewer crimes are registered.
Therefore, it seems that official crime rate statistics are not the best proxy for the
crimes involving corruption.
Another vital discovery of the paper is that integration index has significant
negative impact on CPI. This result is in line with the academic literature examin-
ing cultural corruption determinants and means that more integrated regions ex-
hibit lower values of CPI. However, this is a complex relationship. This is because
highly integrated society may, on the one hand, decrease corruption in its region
resulting in naturally low CPI values; while on the other hand, people may mis-
leadingly undermine the level of corruption in their region due to their moral and
cultural values.
154 Anastasiya Penska

More attention should be paid to the interpretation of spatial analysis conduct-

ed. Global autocorrelation analysis indicated that crime rate has the strongest spa-
tial pattern in the variable list. As the local spatial autocorrelation analysis showed
cluster-like pattern of the regional interrelationship, it is possible that some regions
might get higher corruption rate values as a result of spillover from neighbouring
regions. Therefore, the Ukrainian government should pay more attention to solv-
ing crime issues in the ‘hot points’ of those clusters, mainly Lviv, Zakarpatya,
Ivano-Frankivsk, Donetsk, Odessa and Mykolaiv regions in order to address the
issue more effectively on a larger scale.
Moreover, as GRP growth per capita and real income per capita show the most
influence on CPI, state government might need to pay more attention to the well-
being of the ‘hot-spot’ regions, rather than apply universal policy to all 25 regions.
Indeed, Ukrainian government applies differentiated policy by special develop-
ment policies aimed at less developed regions. However, bearing in mind the spill-
over effect, once corruption is decreased, the government may rethink its strategy
and target the ‘hot-spot’ regions more.
Finally, the spatial dimension of the model allows us to see that integration and
civil organizations have similar global spatial autocorrelation values, and having sig-
nificant impact on CPI, pull it in opposite directions with similar magnitude. While
civil organizations are supposed to help fighting against corruption (as predicted
by theory), it seems that they do the opposite. Main outcome of the study is that there
is vital necessity to influence the way civil organizations in Ukraine operate and en-
sure they are not used for corrupted activities. As civil organization variable showed
one of the highest results in global spatial autocorrelation tests, it is important to pay
more attention to monitoring activities of civil organizations in the regions where
they are mostly clustered. While integration index showed a significant negative im-
pact on corruption, its influence might be lagged as change in mentality is a lengthy
process which is even more complicated in the corrupted environment.

5. Conclusions
In this research the issue of relationship between corruption perception index and
economic, political and social factors for Ukraine was analysed. The question for
the study was selected because of its relevance in today’s development path of
Ukraine and because of the lack of research of spatial dimension in academic lit-
erature. While Coleman (1990) and others find that civil organizations have sig-
nificant negative impact on corruption perception index, this study on Ukrainian
regions challenged this statement. It was shown that civil organizations are an
inefficient tool for fighting corruption in Ukraine, at least the way they are operat-
ing right now.
 Ekonomia nr 42/2015 155

The paper paid special attention to spatial analysis of CPI therefore adding
a new perspective to academic literature. It has been empirically proven that many
of the major economic, political and social variables have spatial dimension in the
case of Ukraine. It should be noted that corruption perception index was used as ae
measure of corruption. This method has brought some limitations to the model as
it is important to bear in mind the subjectivity factor underlying the index. How-
ever, this choice also resulted in more relevant interpretation as it includes real-life
experience of the population.
As a result of the model estimation, the paper verified the originally stated
research hypotheses. It was confirmed that economic growth and real per capita in-
come have negative influence on CPI. The analysed influence of level of crime and
number of MPs originating from a given region did not appear to be statistically
significant. It was further explained that this outcome may result from data im-
perfections. The third hypothesis concerning the impact of civil organizations on
CPI was not supported either. As the paper explains, civil organizations in Ukraine
may serve as tools for money laundering and not strengthening civil society in the
country.
The paper provided solid ground for further analysis in the field. First of all,
when there are more objective statistics available on bribery and corruption crimes,
it would be possible to run the model again and see how the results change. Sec-
ondly, further studies might want to have a closer look at the way population trusts
the government (local and national) as well as examine different spheres where the
population faces corruption the most (universities, healthcare, police etc., see ERA
et al. 2011). This statistics are for now limited and not available for significant time
period as well as not aggregated for the purposes of empirical study but constitute
fruitful soil for further research.
Finally, the paper has come up with some general recommendations that may
be used as guidelines for direction of further development of anti-corruption poli-
cies. While the recommendations majorly address spatial dimension of corruption
problem, they also touch upon the importance of rethinking the role civil organiza-
tions play in fighting corruption.
156 Anastasiya Penska

References
Alam, M. 1995. “A Theory of Limits on Corruption and Some Applications.”
Kyklos 48 (3): 419−435.
Alatas, Syed Hussain. 1986. The Problem of Corruption. Singapore: Times
Books International.
Anselin, Luc and Raymond Florax (eds). 1995. New Directions in Spatial
Econometrics. Berlin: Springer.
Bai, Jai, Seema Jayachandran, Edmund J. Malesky and Benjamin Olken. 2013.
Does Economic Growth Reduce Corruption? Theory and Evidence from
Vietnam. National Bureau of Economic Research Working Paper.
Barro, Robert and Xavier Sala-i-Martin. 1998. Economic Growth. Cambridge:
The MIT Press.
Barro, Robert. 1997. Determinants of Economic Growth: A Cross-Country
Empirical Study. Cambridge: The MIT Press.
Basile, Roberto, Sergio de Nardis and Marianna Mantuano. 2003. Multiple
Regimes in Cross-Region Growth Regressions with Spatial Dependence:
A Parametric and a Semi-parametric Approach. The 43rd European Congress
of the Regional Science Association, Finland. http://www.jyu.fi/ersa2003/
cdrom/papers/4.pdf (accessed 20.01.2014).
Benito, Juan and Roberto Ezcurra. 2005. “Spatial Disparities in Productivity
and Industry Mix: the Case of the European Regions.” European Urban and
Regional Studies 12(2): 177−195.
Bernaldez, Pedro. 2014. “Determinants of Political Corruption:
A ConceptualFramework.” GRAFT and Corruption Research 3(4): 53−79.
Bivard, Roger Gianfranco Piras. 2015. Comparing Implementation of Estimation
Methods for Spatial Econometrics. Journal of Statistical Software 63(18).
Bliss, Christopher and Rafael Di Tella. 1997. “Does Competition Kill
Corruption? ” Journal of Political Economy 105(5): 1001−1023.
Breton, Albert. 1996. Competitive Governments: An Economic Theory of Politics
and Public Finance. Cambridge : Cambridge University Press.
Chenery, Hollis. 1975. “The Structuralist Approach to Development Policy.” The
American Economic Review 65(2): 310−316.
Cliff, Andrew J.K. Ord. 1981. Spatial Processes: Models and Applications.
London : Pion.
Cliff, Andrew and Keith Ord. 1972. “Testing for Spatial Autocorrelation among
regression residuals.” Geographical Analysis 4(3): 267−284.
Coleman, James. 1990. Foundations of Social Theory. Cambridge : Harvard
University Press.
Del Monte, Alfredo and Erasmo Papagni. 2007. “The Determinants of Corruption
in Italy: Regional panel data analysis.” European Journal of Political
Economy 23(2): 379−396.
Diamond, Peter A. 1965. “National Debt in the Neoclassical Growth Model.” The
American Economic Review 55(5): 1126−1150.
 Ekonomia nr 42/2015 157

Drukker, David M., Ingmar R. Prucha and Rafal Raciborski. 2013. “Maximum
Likelihood and Generalized Two-Stage Least-Squared Estimators for
a Spatial-Autoregressive Model with Spatial-Autoregressive Disturbances.”
The Stata Journal 13(2): 221−241.
Drukker, David M. 2009. Analysing Spatial Autoregressive Models Using Stata.
Italian Stata Users Group meeting.
Dzhumashev, Ratbek. 2014. “The Two-Way Relationship Between Government
Spending and Corruption and Its Effect on Economic Growth.” Contemporary
Economic Policy 32(2): 403−419.
European Research Association, USAID, Kyiv International Institute of
Sociology. 2011. Corruption in Ukraine. Comparative Analysis of National
Surveys: 2007–2009, 2011. Kiev: Institute of Sociology.
Fare, Rolf, Shawna Grospkoff, May Norris M. and Zhongyang Zhang. 1994.
“Productivity Growth, Technical Progress, and Efficiency Change in the
Industrialised Countries.” The American Economic Review 84(1): 66−83.
Fearon, James. and David D. Laitin. 1996. “Explaining Interethnic Cooperation.”
The American Political Science Review 90(4): 715−735.
Fiorino, Nadia and Emma Galli. 2010. An Analysis of the Determinants of
Corruption: Evidence from the Italian Regions. POLIS Working Papers, No.
148/171.
Fiorino, Nadia, Emma Galli and Iliaria Petrarca. 2012. “Corruption and Growth:
Evidence from the Italian Regions.” European Journal of Governmentand
Economics 1(2): 127−144.
Glaeser, Edward, Rafael La Porta, Florencio Lopes-de-Silanes and Andrei
Shleifer. 2004. “Do Institutions Cause Growth? ” Journal of Economic
Growth 9(3): 271– 303.
Gupta, Sanjeev, Hamid Davoodi and Rosa Alonso-Terme. 1998. Does Corruption
Affect Income Inequalty and Poverty? International Monetary Fund Working
Paper, No. 98/76.
Harberger, Arnold. 1983. “The Cost-Benefit Approach to the Development
Economics.” World Development 11(10): 863−873.
Hollyer, James R and Leonard Wantchekon. 2011. Corruption in Autocracies.
Social Science Research network. http://dx.doi.org/10.2139/ssrn.1861464.
Huntington, Samuel P. 1968. Political Order in Changing Societies. New Haven:
Yale University Press.
Husted, Bryan. 1999. “Wealth, Culture, and Corruption.” Journal of International
Business Studies 30(2): 339−359.
Jain, A. 2011. Corruption: Theory, Evidence and Policy. Montreal: CESifo DICE
Report.
Johnston, Michael. 1989. “Corruption, Inequality, and Change.” In: Corruption,
Development, and Inequality: Soft touch or Hard Graft, ed. Peter M. Ward,
13−37. London−New York: Routledge.
158 Anastasiya Penska

Jung, Woo S. and Peyton Marshall. 1985. “Inflation and Economic Growth:
Some International Evidence on Structuralist and Distortionist.” Journal of
Money, Credit and Banking 18(2): 227−232.
Kamarianakis, Yiannis and Julie Le Gallo. 2003. The Evolution of Regional
Productivity Disparities in the European Union 1975−2000. Groupement de
Recherches Economiques et Sociales, No. 15.
Kapoor, Mudit and Shamika Ravi. 2009. Determinants of Corruption:
Government Effectiveness vs. Cultural Norms. Social Science Research
Network. http://ssrn.com/abstract=1443617 (accessed 8.06.2015).
Kelejian, Harry and Ingmar R. Prucha. 1999. “A Generalized Moments Estimator
for the Autoregressive Parameter in a Spatial Model.” International Economic
Review 40(2): 509−533.
Koshkina, S. 2015. Майдан. Нерассказанная История, [Maidan. The Untold
Story]. Kyiv: Bright Star Publishing.
Leff, Nathaniel. 1964. “Economic Development Through Bureaucratic
Corruption.” American Behavioral Scientist 8(3): 8−14.
Leibenstein, Harvey. 1968. “Entrepreneurship and Development.” The American
Economic Review 58(2): 72−83.
Linderhof, Vincent, Stijn Reinhard, Eveline van Leeuwen, Martijn Smit,
Peter Nowicki and Rolf Michels. 2013. Spatial Econometric Models for
Evaluating RDP Measures: Analyses for the EU27. Spatial Analysis of Rural
DevelopmentMeasures, Work Package No. 4.
Lipset, Seymour M. 1960. Political Man: The Social Bases of Politics.
Doubleday: Garden City, NY.
Mankiw, Gregory, David Romer and David N. Weil. 1992. “A Contribution to the
Empirics of Economic Growth.” Quarterly Journal of Economics 107(2):
407−437.
Mauro, Paolo. 1995. “Corruption and Growth.” The Quarterly Journal of
Economics 110(3): 681−712.
Melgar, Natalia, Maximo Rossi and Tom. W. Smith. 2009. “The Perception
of Corruption.” International Journal of Public Opinion Research 22(1):
120−131.
Moran, Jon. 1999. “Bribery and Corruption: The OECD Convention on
Combating the Bribery of Foreign Public Officials in International Business
Transactions.” Business Ethics: A European Review 8(3): 141−150.
Myrdal, Gunnar. 1968. The Asian Drama. Harmondsworth: The Penguin Press.
Neutze, Jand and Adrian Karatnycky. 2007. Corruption, Democracy and
Investment in Ukraine. Washington: The Atlantic Council of the United States.
North, Douglass C. 1971. “Institutional Change and Economic Growth.” Journal
of Economic History 31(1): 118−125.
Olson, Mancur. 1996. “Big Bills Left on the Sidewalk: Why Some Nations Are
Rich, And Others Poor.” Journal of Economic Perspectives 10(2): 3−24.
Ord, Keith. 1975. “Estimation Methods for Models of Spatial Interaction.”
Journal of the American Statistical Association 70(349): 120−126.
 Ekonomia nr 42/2015 159

Pigliaru, Francesco. 2009. Persistent regional Gaps and the Role of Social
Capital: Hints from the Italian Mezzogiorno’s Case. Centre for North South
Economic Research Working Papers No. 04.
Pisati, Maurizio. 2012. Exploratory Spatial Data Analysis Using Stata. Social
Science Research Center. http://www.stata.com/meeting/germany12/abstracts/
desug12_pisati.pdf (accessed 1.12.2015).
Putnam, Robert. 1993. Making Democracy Work: Civic Traditions in Modern
Italy. Princeton : Princeton University Press.
Romer, David. 2001. Advanced Macroeconomics. New York : McGraw-Hill.
Sandholtz, Wayne and William Koetzle. 2000. „Accounting for Corruption:
Economic Structure, Democracy and Trade.” International Studies Quarterly
44(1): 31−50.
Senior, Ian. 2006. The World’s Big C: Cases, Causes, Consequences, Cures.
London: Institute of Economic Affaires.
Serra, Danila. 2004. “Empirical Determinants of Corruption: A Sensivity
Analysis.” Public Choice 126(1/2): 225−256.
Shabbir, Ghulam and Mumtaz Anwar. 2007. “Determinants of Corruption in
Developing Countries.” The Pakistan Development Review 46(4): 751−764.
Shkurpat, Oleksandr. 2006. Regional Labor Productivity Disparities inUkraine:
Main Causes and Spatial Patterns. Kyiv: Kyiv School of Economics.
Solow, Robert M. 1956. “A Contribution to the Theory of Economic Growth.”
Quaterly Journal of Economics 70(1): 65−94.
Sonin, Konstantin. 2000. Inequality, Property Rights Protection and Economic
Growth in Transition Economies. EERC Working Paper No. 2K/02E.
Spector, Betram, Svetlana Winbourne, Jerry O’Brien and Eric Rudenshiold.
2006. Corruption Assessment: Ukraine. Washington, D.C.: Management
Systems International Corporate Offices.
Tanzi, Vito and Hamid Davoodi. 1998. “Corruption, Public Investment and
Growth.” In: The Welfare State, Public Investment and Growth, eds. Hirofumi
Shibata, Toshihiro Ihori, 41−60. Tokyo: Springer
Tokarchuk, T. 2006. Determinants of Economic Growth in Ukraine.Working
Paper, No. 12 (1).
Transparency International. 2014. Corruption Perception Index. Transparency
International. Annual.
Treisman, Daniel. 2000. “The Causes of Corruption: A Cross-National Study.”
Journal of Public Economics 76(3): 399−457.
Tsyrennikov, V. 2002. Determinants of Regional Growth. Kyiv: Kyiv School of
Economics.
Vyshnyak, Oleksandr. 2009. Мовна Ситуація в Україні: Динаміка,
Проблеми, Перспективи (Соціологічний аналіз) [Linguistic Situation
and Status in Ukriane: Dynamics, Problems, Perspectives (Sociological
Analysis)]. Kyiv: Institute of Sociology of Ukraine. In Ukrainian.
160 Anastasiya Penska

Wines, William and Nancy K. Napier. 1992. “Toward an understanding of

Cross-Cultural Ethics: A Tentative Model.” Journal of Business Ethics 11(11):
831–841.
Комітет Державного Бюджету України [Ukrainian Public Bugdet
Committee]. 2013. Висновки щодо Державного Бюджету України,
[Conclusions Regarding Allocation of Public Budget]. Kyiv: Ukrainian Public
Bugdet Committee. In Ukrainian.
Статистичний Комітет України [Ukrainian Statistical Bureau]. 2005−2011.
Освіта та Інновації в Україні, [Educational and Innovational Activity in
Ukraine]. Kyiv: Ukrainian Statistical Bureau. In Ukrainian.
—. 2005‒2011. Регіональний Людський Розвиток, [Regional Human
Development]. Kyiv: Ukrainian Statistical Bureau, 2005−2011. In Ukrainian.
—. 2005‒2011. Статистичний Щорічник України, [Annual Statistical
Yearbook of Ukraine]. Kyiv: Ukrainian Statistical Bureau, 2005−2011. In
Ukrainian.
—. 2005‒2011. Щорічник Регіонів України, [Annual Yearbook of Ukrainian
Regions]. Kyiv: Ukrainian Statistical Bureau. In Ukrainian.

Index of abbreviations
CPI – Corruption Perception Index
GRP – Gross Regional Product
MP – Member of Parliament
UAH – Ukrainian Hryvnia
USB – Ukrainian Statistical Bureau
ERA – European Research Association
2SLS (IV) – Two-Stage Least Squares estimation with Instrumental Variables